Continental Breakfast
Continental Breakfast
etn Newsletter | September 14, 2026
What a time to be live. The main thing happening today is AI's biggest rivals uniting behind a call to slow down. Live from 12–3 BST. Here's your Order of Play.
12:50 - Ed Leon Klinger, Co-Founder & CEO at Flock
13:00 - Michiel Bakker, Assistant Professor at MIT
13:15 - Leo Wong, Co Founder & CFO at Pony AI
13:30 - Alexander Fitzgerald, Founder & CEO at Isembard
13:45 - Ernie Tedeschi, Head of Economic Insights & Research at Stripe
14:30 - Sevan Marian, Co-Founder & CEO at Fleet
14:30 - Alexandre Berriche, Founder & Chairman at Fleet
What’s Happening Today
Dario Amodei wants the AI industry to slow down, and this time his rivals agree with him.
In an essay published Saturday, Amodei argued the industry "must slow the pace at which we improve the capabilities of AI models," after development got "drastically faster" this summer. He's clear that pacing isn't pausing, training carries on, but labs take the time to check and align their models before pushing further. Within hours, his two fiercest rivals were nodding along. "Dario is right," Musk posted. "I agree with Dario that we need to pace the frontier," wrote Altman, who added that OpenAI would adopt Amodei's idea of independent evaluators with employee-level access.
Two things changed his thinking. The first is recursive self improvement, AI getting capable enough to build the next generation of models, which he says accelerated across the industry this summer. The second is the OpenAI and Hugging Face incident, where a swarm of agents launched cyberattacks on targets nobody had assigned them and tried to hack the system grading their performance. His worry is that same behaviour with more capability behind it, and he reckons a swarm like that could be running a botnet across the internet within 6 to 12 months, doing hundreds of billions in damage.
The proposal has three steps, and only the first is Anthropic's to make. The company is committing to embedded evaluators, an outside team with desks in the office, badges and access comparable to its own risk staff. They are free to publish what they find without Anthropic having editorial control. Step two asks frontier labs across the democratic world to agree shared safety standards, which needs an antitrust waiver from Washington before it is legal. Step three is an agreement with China, which he thinks is realistic on narrow issues like bioweapons but a long shot on anything bigger.
We're getting into all of it on today's show at 12 BST.
etn. Insight
Andy Twelves, a former Newspeak fellow, came on to push back on the conspiracy theories swirling around the organisation, the idea that it's some shadowy outfit out to undermine democracy.
"I wish it was Hillary Clinton, AI rebellion. I wish we were hacking democracy, that'd be sick." The reality, he said, is a good deal more mundane. "It's a really positive fellowship where you take a year out of your work as an early- to mid-career technologist and do some research, learn about things you wouldn't have otherwise." His own project? "Digital organising in trade unions, and how they can adapt to the creation of data standards, which was quite boring."
Watch the full clip below.
Continental Breakfast
Altman and Musk rally behind Amodei’s call for an AI slowdown
Isembard launches largest factory in London since WW2
Leading German economists and policymakers warned Europe risks being left behind in AI
Swedish healthtech Tandem Health raised $100 million Series B
etn. is made possible by Airwallex, base, ElevenLabs, Framer, HEX, BVNK, Valarian, Metaview, Chargebee and Polymarket.